RYS-Phi-3-medium-4k-instruct — benchmark results
dnhkng's RYS-upscaled variant of Microsoft's Phi-3-medium-4k-instruct, enlarged to ~18B params via an experimental layer-expansion technique. Provider: Other. Released 2024-08-06. Access: Open.
Unified ELO 1466 ± 44, rank #928 of 1776 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - GPQA | 13.98 | Score | 89.4 |
| Open LLM Leaderboard - MMLU-Pro | 42.74 | Score | 86.8 |
| Open LLM Leaderboard - BBH | 46.75 | Score | 86.7 |
| Open LLM Leaderboard - MATH Level 5 | 16.09 | Score | 64 |
| Open LLM Leaderboard - MuSR | 11.09 | Score | 57.2 |
| Open LLM Leaderboard - IFEval | 43.91 | Score | 47.1 |
| Open PL LLM - Generative | 30.59 | Average Generative Score (%) | 29.7 |
| Open PL LLM Leaderboard | 22.04 | Average Score (%) | 26.9 |
| Open PL LLM - Multiple Choice | 10.93 | Average Multiple-Choice Score (%) | 25.5 |
| Open PL LLM - RAG | 37.23 | Average RAG Score (%) | 25.2 |
Interactive version: theaggregate.ai/model?slug=rys-phi-3-medium-4k-instruct · How the rankings work · Data refreshed daily, snapshot 2026-07-22.